Comparison and Evaluation of the Molecular Typing Methods for Toxigenic Vibrio cholerae in Southwest China

Vibrio cholerae O1 strains taken from the repository of Yunnan province, southwest China, were abundant and special. We selected 70 typical toxigenic V. cholerae (69 O1 and one O139 serogroup strains) isolated from Yunnan province, performed the pulsed field gel electrophoresis (PFGE), multilocus sequence typing (MLST), and MLST of virulence gene (V-MLST) methods, and evaluated the resolution abilities for typing methods. The ctxB subunit sequence analysis for all strains have shown that cholera between 1986 and 1995 was associated with mixed infections with El Tor and El Tor variants, while infections after 1996 were all caused by El Tor variant strains. Seventy V. cholerae obtained 50 PFGE patterns, with a high resolution. The strains could be divided into three groups with predominance of strains isolated during 1980s, 1990s, and 2000s, respectively, showing a good consistency with the epidemiological investigation. We also evaluated two MLST method for V. cholerae, one was used seven housekeeping genes (adk, gyrB, metE, pntA, mdh, purM, and pyrC), and all the isolates belonged to ST69; another was used nine housekeeping genes (cat, chi, dnaE, gyrB, lap, pgm, recA, rstA, and gmd). A total of seven sequence types (STs) were found by using this method for all the strains; among them, rstA gene had five alleles, recA and gmd have two alleles, and others had only one allele. The virulence gene sequence typing method (ctxAB, tcpA, and toxR) showed that 70 strains were divided into nine STs; among them, tcpA gene had six alleles, toxR had five alleles, while ctxAB was identical for all the strains. The latter two sequences based typing methods also had consistency with epidemiology of the strains. PFGE had a higher resolution ability compared with the sequence based typing method, and MLST used seven housekeeping genes showed the lower resolution power than nine housekeeping genes and virulence genes methods. These two sequence typing methods could distinguish some epidemiological special strains in local area.


INTRODUCTION
Vibrio cholerae is a Gram-negative intestinal pathogen, causing serious human diarrhea, mainly distributed in southern Asia, parts of Africa, Latin America, and other regions (Heidelberg et al., 2000;Morris, 2011). Toxigenic V. cholerae is the strain carrying cholera toxin (CT), and mainly refers to O1 and O139 serogroup (Faruque et al., 1998;Nair et al., 2006). However, non-O1/non-O139 V. cholerae is not carrying CT, and can only cause mild diarrhea diseases (Singh et al., 2001). Therefore, the prevention and control of toxic strains are more important for humans. In China, cholera was considered to be one of the most serious infectious diseases, although the incidence rate has been maintained at a relatively low level in recent years, the epidemic or outbreak still existed in few areas (Gu et al., 2014). It is very important to perform the molecular typing research for toxigenic V. cholerae and clarify the variation and changes of bacteria.
At present, the majority molecular typing methods of V. cholerae comprised of pulsed field gel electrophoresis (PFGE), multilocus sequence typing (MLST), MLVA (multiple-locus variable number tandem repeat analysis), or genome sequencing (Karaolis et al., 2001;O'Shea et al., 2004a,b;Danin-Poleg et al., 2007;Grim et al., 2010;Taviani et al., 2010;Okada et al., 2012;Sealfon et al., 2012;Tran et al., 2012). PFGE is considered to have highly discrimination efficiency, and commonly used in the epidemiological or outbreak investigation. Two MLST typing methods have been reported, one was used seven housekeeping genes for adk, gyrB, metE, pntA, mdh, purM, and pyrC, established by Octavia 1 (Octavia et al., 2013). This method has established the database, and researchers in different countries could submit and compare their results. Another was used nine housekeeping genes for cat, chi, dnaE, gyrB, lap, pgm, recA, rstA, and gmd. This method was developed by Garg et al. (2003), several studies have used this method to perform their researches, showing a good discriminatory power (Bhattacharya et al., 2006;Ang et al., 2010). However, this method has not yet established a public database. Researchers from different regions were unable to exchange and share their data. In addition, some studies performed the molecular typing researches by using virulence genes; the results also had effective resolving abilities (Rivera et al., 2001). Up to present, there was no systemic evaluation for molecular typing methods of toxigenic V. cholerae, especially for two MLST methods mentioned above. The applicability of different typing methods was still unknown.
Yunnan located in southwest China, bordering Myanmar, Vietnam, and Laos, has an extended frontier. V. cholerae resources here were abundant and special, indicated that the cholera was endemic in these regions. Although cholera cases were seldom found in recent years, the imported strains from neighboring countries still existed (Liao et al., 2016). It was very important to find the epidemic consistency of cholera by molecular typing methods. In this study, we selected 70 typical toxigenic V. cholerae isolated from different areas and years in 1 http://pubmlst.org/vcholerae Yunnan province, performed the PFGE, two MLST typing, and MLST of virulence gene (V-MLST) methods, and compared the distinguish ability for different molecular typing methods in local epidemic area.

Strains
Seventy V. cholerae strains (already-existing collections) were isolated from different regions, years, and sources in Yunnan province between 1986 and 2012. Sixty-nine strains were O1 serogroup, included 43 Ogawa and 26 Inaba serotype, and one O139 serogroup isolates (we only have three O139 serogroup strains, and selected one as the representative for the study purpose). Fifty-four strains were isolated from the feces samples of patients, 11 from water samples, and five from the external environment (surface of objects), as shown in Table 1.

PCR Detection of Virulence Genes and ctxB Sequencing
Genomic DNA was extracted from each isolate using a DNA extraction kit (Tiangen, Beijing) according to the manufacturers' instructions. The virulence genes for ctxAB, ompU, ace, zot, toxR, rtxC, and CTX phage rstR (Classical/El Tor) and tcpA (Classical/El Tor) were amplified using Taq premix (TaKaRa, Japan), the primers and amplification procedures were as described previously (Chow et al., 2001;Singh et al., 2001Singh et al., , 2002O'Shea et al., 2004b). All of the strains were sequenced for ctxB gene subunit to further identify the characters of the CTX phage, Taq premix (TaKaRa, Japan) was used as described above, and amplification processes were performed as previously described (Goel et al., 2010). The amplification products were sent for bidirectional sequencing (TaKaRa, Japan), and the results were analyzed using DNAStar (DNASTAR, Inc., United States) and MEGA 4 software (Tamura et al., 2007). The ctxB sequences of N16961 of El Tor V. cholerae (GenBank: NC-002505) and O395 Classical strain (GenBank: NC-012582) were used as the standards for comparison.

Pulsed Field Gel Electrophoresis
Pulsed field gel electrophoresis was performed based on the PulseNet protocol for V. cholerae and procedures described previously (Gu et al., 2014). The enzyme digestion for each plug was NotI 40U at 37 • C for 4 h. The CHEF-Mapper (Bio-Rad) was used for electrophoresis, and the pulse time ranged from 1 to 20 s for 13 h, and 20 to 25 s for 6 h. The gel was stained using Gel Red (Biotium) and visualized using the gel imaging system (Bio-Rad, Gel Doc XR). PFGE patterns were analyzed with BioNumerics version 6.6 (Applied Maths, Belgium), and a dendrogram was produced using the Dice coefficient and un-weighted pair group method with arithmetic mean algorithm (UPGMA). A pairwise distance matrix was also created.
ET, El Tor biotype; CL, Classical biotype. The numbers in the bracket represented the strains possessed the ET or ET, CL alleles in different counties and years, no bracket represented all the isolates had the same allele.

Nine Housekeeping Genes
PCR amplification was made following the published work (Garg et al., 2003) targeting the genes cat, chi, dnaE, gyrB, lap, pgm, recA, rstA, and gmd. The primers were shown in

Virulent Genes of MLST
We designed the ctxAB, tcpA, and toxR genes primers ( Table 2) by using Clone Manager Professional 8.0 software (Scientific & Educational) as well, V. cholerae reference strain N16961 was also used. The reaction system and amplification procedure were identical as mentioned above. Amplified products were sent to bidirectional sequencing (TaKaRa, Japan).

Data Analysis
All the sequencing results were assembled by DNAStar 6.0 software (DNASTAR, Inc., United States), compared and aligned by MEGA 4.0 (Tamura et al., 2007). The seven housekeeping genes sequences were submitted to the public database (see text footnote 1), the alleles of different genes and sequence types (STs) were obtained. The sequence alignments were performed for nine housekeeping and virulence genes, when a new sequence appeared; we gave a new allele for each gene, and finally got the STs by permutation and combination of the nine genes or virulent genes. The minimum

Nucleotide Sequence Accession Numbers
All the genes of different sequences were deposited in the GenBank with the accession numbers: KX960341 to KX960367.

Ethics Approval Statement
The human sample collection and detection protocols were carried out in accordance with relevant guidelines and regulations approved by Ethical Committee of Yunnan Provincial Centre for Disease Control and Prevention. All experimental procedures were approved by the Ethics Review Committee [Institutional Review Board (IRB)] of Yunnan Provincial Centre for Disease Control and Prevention. All adult subjects provided informed consent, and a parent or guardian of any child participant provided informed consent on their behalf. The informed consents were oral for all the participants, because the samples were too large; we could not get all the written ones. All samples collections and experimental procedures were approved by the Ethics Review Committee, according to Chinese ethics laws and regulations. The anonymization strategy was used for the human sample collection and detection protocols used in this study. The details of patients, such as name, address, age, and sex were anonymous, and we just defined the numbers of patients or samples.

PCR Test for Virulence Genes and ctxB Sequencing
The ctxAB, ompU, ace, zot, toxR, and rtxC for all of the isolates were positive; tcpA ElTor was positive for all of the isolates as well, while tcpA Classical was negative. For the rstR, most of the strains carried rstR ElTor and rstR Classical ; however, some of the strains possessed only rstR ElTor . The ctxB subunit showed mixed infection with El Tor type and El Tor variant strains before 1995; after 1996 all of the isolates harbored the ctxB Classical except one O139 V. cholerae that possessed ctxB El Tor (Table 1).

PFGE Results
Seventy toxigenic V. cholerae obtained 50 PFGE patterns, with a high resolution. The clusters could be divided into three groups, named as A-1, A-2, and B with predominance of strains isolated during 1980s, 1990s, and 2000s, respectively (Figure 1) in 1989and 1991, Yuanmou in 1994and 1997, Ruili in 1995and 1998, Yongshan in 1996, Yanshan, Guangnan, and Mangshi in 1998. Compared the PFGE result with our previous study (Liao et al., 2016), we found that PFGE had highly discrimination power with whole genomic sequencing method, since the imported strain YN2011QXL (YN2011004) was separated from other three V. cholerae in our previous work used genomic sequencing. And in this study, YN2011QXL was also clustered to different groups with other V. cholerae.

MLST and V-MLST Results
The MLST results used seven housekeeping genes showed that all the strains belonged to ST69. adk allele was 7, gyrB was 11, metE 37, pntA 12, mdh 4, purM 1, and pyrC 20. The results had no relations with isolated areas or years of the strains (Figure 2A). Nine housekeeping genes were arranged and combined to produce seven different STs, named as ST1-ST7, as shown in Figure 2B. Three imported strains from Myanmar after 2011 formed their own ST (Figure 2B, blue area). Three virulence genes were analyzed and produced nine STs, named as ST1-ST9, as shown in Figure 2C. The imported strains also formed their own STs, while YN2011QXL and other two strains were divided into different types. The latter two sequences based typing methods had consistency with epidemiology of the strains. Seven STs were found for all the isolates used nine housekeeping genes method, rstA gene had five alleles (YN2011DW, YN2011QXL, and YN2012SW; YN91226; YN95173; YN96022; and other strains), recA had two alleles (YN86041 and other strains), gmd had two alleles (YN98336 and other strains). Other six housekeeping genes had only one allele, respectively. For rstA gene, YN91226 mutated at position 505 nt; YN2011DW mutated at 453, 459, and 468 nt; YN95173 mutated at 453 nt; and YN96022 mutated at 468 nt, as Figure 3A shown. For recA gene, YN86041 inserted a "T" at position 105 nt ( Figure 3B). For gmd gene, YN98336 mutated at 11,17,20,22,34,36,42,43,56, and 104 nt position ( Figure 3C).
All the sequence alignments results for genotyping methods were shown in Supplementary Material.

Comparison the Molecular Typing Methods
Compared the sequence based typing methods in this study, genotyping of seven housekeeping genes was unable to distinguish between strains from different epidemiological resources; genotyping of nine housekeeping genes divided 70 strains into seven STs, and the different epidemiological resources of isolates were distinguished by this method; genotyping of three virulence genes had the similar discriminatory power with nine FIGURE 2 | The minimum spanning tree of three sequence based genotyping methods. (A) Seven housekeeping genes: all the strains belonged to ST69; (B) nine housekeeping genes: seven STs were found by using this method, named as ST1-ST7. The typical isolates were shown below each STs; and (C) virulence genes: nine STs were found by using this method, named as ST1-ST9. The typical isolates were shown below each STs. housekeeping genes. However, the discriminatory ability based on sequence typing methods was lower than PFGE in the local epidemic areas. For example, the cholera epidemic happened in 1986 of Gengma County (A-1 group); 10 patterns were found among 11 strains, while only two STs were found used nine housekeeping genes method (ST1 and ST2); and three STs were identified used virulence genes method (ST1, ST2, and ST7).

DISCUSSION
Pulsed field gel electrophoresis is considered to be the "golden standard" for pathogen molecular typing techniques, showing the highly resolution power, frequently used for outbreak investigation and traceability analysis. MLST is often used for analysis the long-term variability and changes of strains. The purpose of our study was evaluated different molecular typing methods for toxigenic V. cholerae in Yunnan province. From our previous works (Gu et al., 2014;Liao et al., 2016), V. cholerae in Yunnan province, southwest China had similar homology with strains from other part of China, or even some southeast Asia countries. Therefore, the isolates used in this study could reflect the general V. cholerae distributions characteristics in China or southeast Asia, and have enough representatives of the bacteria.
At present, PFGE have the standard experimental procedure, the data could be exchanged and analyzed between different laboratories. Two MLST methods have been reported, Garg et al. (2003) analyzed 96 O139 strains used nine housekeeping genes in 2003, and they found 64 new alleles in 51 STs. Several studies have used this method to perform their researches, for example, Nguyen et al. (2009) performed the MLST method for cholera outbreak in Vietnam in 2009 by using nine housekeeping genes, and all the strains had the same ST with N16961. Lee et al. (2006) analyzed the Mozambique V. cholerae by using nine genetic loci showed that the Mozambique isolates have the same ST as O1 El Tor N16961. Kotetishvili et al. (2003) used 22 V. cholerae isolates to perform the PFGE and MLST by using three housekeeping genes, gyrB, pgm, and recA; sequence data were also obtained for the virulence-associated genes tcpA, ctxA, and ctxB. Their results showed that MLST had better discriminatory ability than PFGE; On MLST analysis, there was clear clustering of epidemic serogroups; much greater diversity was seen among tcpA and ctxAB positive V. cholerae strains from others, non-epidemic serogroups, with a number of tcpA and ctxAB alleles identified. However, this method has not established public database, and its application was limited. Octavia et al. (2013) developed a new MLST method for V. cholerae in 2013; they found a total of 77 isolates were divided into 66 STs, including 55 non-O1/non-O139 strains. While, in this study, 70 toxigenic strains had only one ST, no correction with epidemiological information could be found with the typing results, we considered this method was more suitable for genomic diversity of non-O1/non-O139 V. cholerae. In fact, in our study, PFGE had higher discriminatory power than all sequence based typing method. The cholera epidemic happened in 1986 of Gengma County (A-1 group); 10 patterns were found among 11 strains, while only two STs were found used nine housekeeping genes method (ST1 and ST2); and three STs were identified used virulence genes method (ST1, ST2, and ST7). Therefore, we considered that PFGE was more suitable for molecular typing in cholera epidemic local area.
In Boyd and Waldor (2002) study, genetic variation at the tcpA locus in toxigenic isolates of V. cholerae was investigated; the results showed tcpA sequences were far more diverse than other loci. This diversity was a reflection of diversifying selection in adaptation to the host immune response. Therefore, we selected three major virulence genes of V. cholerae to perform the molecular typing analysis. Its discriminatory ability was similar with nine housekeeping genes method, and the tcpA gene discriminatory effect was the best compared with ctxAB and toxR.
Compared the sequence based typing methods in this study, genotyping of seven housekeeping genes was unable to distinguish between strains from different epidemiological resources; genotyping of nine housekeeping genes divided 70 strains into seven STs, and the different epidemiological resources of isolates were distinguished by this method; genotyping of three virulence genes had the similar discriminatory power with nine housekeeping genes. However, the discriminatory ability based on sequence typing methods was lower than PFGE in the local epidemic areas.

AUTHOR CONTRIBUTIONS
BK, HJ, and WG designed the work. FL, MC, BP, XF, and WX did the experiments. ZM and WG analyzed the data. ZM drafted the work. BK and HJ revised it critically for important intellectual content.